The purpose of the proposed research is to understand how biologically plausible mechanisms lead to controlled behavior. It is commonly acknowledged that there are typically many different ways to perform cognitive tasks, and further, that individual differences in personal strategies are a potential problem for experimental cognitive neuroscientists. However, there is very little theoretical development of why particular strategies are selected, or even how to address this issue. The proposed research uses computational simulations to investigate the trade-offs between different implemented strategies in the performance of a complex sequential working memory task. Further, predictions from such simulations are tested in subsequent empirical work. The trade-offs are investigated in the context of how quickly the different strategies learn to perform the task, the internal representations that they use to perform the task, and asymptotic performance. These studies will provide a formal basis for understanding how task demands influence the selection of different strategies, and they will provide experimental cognitive neuroscientists with additional insights into how to develop new paradigms that constrain the strategies that participants use. In addition to developing scientific tools, such an understanding will facilitate the development of behavioral therapies for patients with deficits in control, such as older adults and patients with schizophrenia. This project serves public health by developing a more comprehensive understanding of how humans perform difficult tasks that require control. By understanding why participants select the strategies they use to perform such tasks, researchers can develop new therapies that encourage the most efficient alternate strategies when the optimal ones are incapacitated through trauma, aging, or pathology. [unreadable] [unreadable] [unreadable]